Classification method and apparatus

Abstract

Processing data derived from currency items includes measuring samples of currency items of classes to produce feature vectors in a first space and mapping the feature vectors to a second space in which there is a clearer separation of the classes.

1. A method of processing data derived from currency items, the method comprising:
 measuring samples of currency items of classes to produce feature vectors in a first space; and  
 mapping the feature vectors to a second space in which there is a clearer separation of the classes than in the first space.  
 
   
   
     2. A method as claimed in  claim 1  wherein the second space is determined using PCA or LDA. 
   
   
     3. A method as claimed in  claim 2  wherein data in the first space is mapped to a third space using a non-linear mapping, and PCA or LDA is performed on the image of the data in the third space. 
   
   
     4. A method as claimed in  claim 1  further comprising deriving a separating function separating the classes. 
   
   
     5. A method of classifying an item of currency comprising measuring the currency item to derive a feature vector and determining the class of the currency item using the result of a method as claimed in  claim 1 . 
   
   
     6. A method of programming a currency validator comprising:
 performing a method according to  claim 1  using a plurality of samples of currency items; and  
 using a result from said performing to program the validator.  
 
   
   
     7. An apparatus for classifying items of currency comprising sensors to sense features of a currency item, a processor to generate a feature vector from the sensed values and a classifier to classify currency items using a result of a method as claimed in  claim 1 . 
   
   
     8. An apparatus for classifying items of currency comprising a classifier to process feature vector data derived from currency items wherein the classification is equivalent to processing the input data, which is in a first space, in a second space where there is clearer separation of the classes than in the first space. 
   
   
     9. An apparatus as claimed in  claim 8  wherein the second space is determined using PDA or LDA. 
   
   
     10. An apparatus as claimed in  claim 8  wherein the second space is determined using PDA or LDA, in the image of the input space under a non-linear mapping.